Bearing Monitoring: How to Track and Protect Bearings
Key Takeaways
- Bearings are the most commonly monitored component in rotating equipment because bearing failure is the leading cause of motor and gearbox breakdowns.
- Bearing degradation follows a predictable four-stage model, from early sub-surface cracking detectable only by ultrasound to near-catastrophic failure detectable by all methods.
- Selecting the right monitoring technique depends on machine speed, criticality, failure stage, and available lead time before intervention is required.
What Is Bearing Monitoring?
Bearing monitoring is a core discipline within asset condition monitoring. Its purpose is to close the gap between scheduled maintenance intervals and actual bearing health by providing continuous or periodic evidence of what is happening inside the machine.
Unlike calendar-based maintenance, bearing monitoring ties intervention to real equipment condition. A bearing showing no detectable deterioration can run beyond its design interval. One showing Stage 2 degradation gets maintenance scheduled within weeks, before the fault reaches Stage 3 and unpredictable failure becomes likely.
Why Bearings Are the Most Monitored Component in Rotating Equipment
Bearings fail more frequently than any other component in motors, pumps, compressors, fans, and gearboxes. Industry data consistently shows that bearing failure accounts for 40 to 50 percent of all rotating machinery breakdowns. The consequences range from production stoppages and costly repairs to secondary damage on shafts, housings, and connected equipment.
Three factors make bearings particularly prone to failure:
- High duty cycle. Bearings operate continuously under load, often for years without inspection.
- Sensitivity to operating conditions. Small changes in lubrication quality, alignment, or load distribution can dramatically shorten bearing life.
- Hidden failure initiation. Sub-surface fatigue cracks develop well before they produce detectable noise, temperature rise, or vibration. By the time an operator notices something wrong, significant damage has often already occurred.
These characteristics make bearings the highest-priority target for predictive maintenance programs. Early detection translates directly into planned repairs rather than emergency shutdowns.
The Four-Stage Bearing Failure Model
Bearing degradation follows a well-established progression. Understanding where a bearing sits in this model determines which monitoring technique will detect it and how much lead time is available for intervention.
Stage 1: Sub-Surface Stress (Ultrasound Range)
Sub-surface fatigue cracks form in the raceway or rolling elements. The bearing may appear and sound completely normal. Standard vibration measurements show no anomaly. Only high-frequency ultrasound analysis, operating in the 20 to 100 kHz range, can detect the stress waves emitted by micro-crack propagation at this stage. Lead time from Stage 1 to Stage 4 on a critical machine can range from weeks to many months depending on load and speed.
Stage 2: High-Frequency Vibration Spikes
Defects become large enough to produce impulse forces when rolling elements pass over them. These impulses appear as energy spikes in the high-frequency vibration spectrum (typically 500 Hz to 2 kHz). The bearing still runs smoothly to the touch and ear. Envelope analysis and high-frequency band kurtosis are the most sensitive measurement techniques at this stage. The P-F curve shows maximum lead time at Stage 2: detecting faults here provides the most planning time before failure.
Stage 3: Visible Sidebands and Rising Overall Levels
Fault frequencies (ball pass frequency, inner race, outer race) become clearly visible in the vibration spectrum as sidebands around rotational harmonics. Overall vibration levels begin to rise measurably. The bearing generates audible noise. Temperature may start to increase. At Stage 3, the machine should already be in a planned maintenance window, with parts on order and a repair date confirmed.
Stage 4: Near-Catastrophic Failure
Structural integrity is severely compromised. Vibration levels spike erratically. Temperature rises rapidly. Metal-to-metal contact begins. The bearing can seize or shed material within hours. Any machine found at Stage 4 through vibration monitoring should be evaluated for immediate shutdown.
Bearing Monitoring Techniques
No single technique monitors all stages of bearing failure equally well. Effective programs combine methods based on machine criticality, speed, and available resources.
Vibration Analysis
Vibration analysis is the most widely used bearing monitoring method. Accelerometers mounted on bearing housings measure the velocity and acceleration of structural vibration. The resulting signal is processed in both time and frequency domains to extract fault indicators.
Key parameters used in vibration-based bearing monitoring:
- RMS velocity (mm/s). The root mean square of velocity over a broad frequency range. Used for overall severity assessment and ISO 10816/20816 severity zone classification.
- Peak acceleration (g). The highest instantaneous acceleration value. Sensitive to impact events associated with spalling and cracking.
- Kurtosis. A statistical measure of signal impulsiveness. Healthy bearings produce kurtosis values near 3. Values above 6 indicate significant defect-related impacting. Kurtosis is most useful at Stages 1 and 2, before overall levels rise.
- Crest factor. The ratio of peak amplitude to RMS amplitude. Rising crest factor indicates increasing impulsiveness relative to background vibration, an early indicator of bearing damage.
- Bearing fault frequencies. Calculated from bearing geometry (ball pass frequency outer race, ball pass frequency inner race, ball spin frequency, fundamental train frequency). These frequencies and their harmonics appear in the spectrum as defect-specific signatures.
Ultrasound Analysis
Ultrasound analysis detects high-frequency acoustic emissions produced by stress waves, friction, and turbulent fluid flow. It is the only widely available technique capable of detecting Stage 1 bearing defects. It is also the preferred method for very slow-speed bearings (below 100 RPM) where conventional vibration analysis lacks the sensitivity to detect early faults.
Ultrasound is also the most practical tool for real-time lubrication assessment. As lubricant film breaks down, friction increases and ultrasonic decibel levels rise. This allows technicians to detect lubrication deficiency before it generates heat or surface damage, and to verify that relubrication has restored adequate film.
Temperature Monitoring
Thermal monitoring uses contact sensors (RTDs, thermocouples) or infrared thermography to track bearing housing temperature. Temperature rise is a relatively late-stage indicator. By the time a bearing generates excess heat, the fault is typically Stage 3 or early Stage 4. Temperature monitoring is most valuable as a continuous safety trip parameter on critical machines and as a complement to vibration analysis for confirming fault severity. Normal operating bearing temperatures run between 50 degrees C and 90 degrees C depending on machine type and ambient conditions; a rise of 10 to 15 degrees C above established baseline warrants investigation.
Oil Analysis
For bearings in oil-bath, circulating oil, or oil-mist lubrication systems, oil analysis provides indirect evidence of bearing condition through the particles and chemical markers present in the lubricant. Wear particle analysis reveals the size, shape, and composition of metallic debris, which distinguishes normal wear from abrasive cutting, fatigue spalling, and corrosive attack. Spectrometric analysis quantifies elemental concentrations. Particle count and ISO cleanliness codes track contamination levels. Oil analysis complements vibration and ultrasound by identifying the root cause of degradation, particularly lubrication contamination and breakdown.
Visual and Manual Inspection
Direct visual inspection during scheduled downtime remains part of a complete bearing monitoring program. Technicians assess raceway and rolling element surface condition, cage integrity, and lubricant appearance. Visual inspection confirms findings from instrument-based monitoring and provides ground truth for calibrating detection thresholds. It is not a substitute for instrument-based methods, since visual inspection can only find faults that have already produced macro-scale damage.
Monitoring Technique Comparison
| Technique | Failure Stage Detected | Best Application | Limitation |
|---|---|---|---|
| Ultrasound | Stage 1, 2 | Early detection, slow-speed bearings, lubrication assessment | Requires close proximity; less effective through complex structures |
| Vibration analysis | Stage 2, 3, 4 | Fault identification, severity trending, ISO classification | Limited sensitivity below ~100 RPM; misses Stage 1 |
| Temperature monitoring | Stage 3, 4 | Safety trips, severity confirmation on critical machines | Late indicator; insufficient lead time for planned repair |
| Oil analysis | Stage 1, 2, 3 | Root cause identification, lubrication condition, contamination | Only applicable to oil-lubricated systems; lab turnaround adds latency |
| Visual inspection | Stage 3, 4 | Confirmation during scheduled downtime, root cause documentation | Requires shutdown; misses sub-surface and early-stage damage |
Vibration Parameters Explained
Practitioners new to bearing monitoring often encounter vibration parameters without clear guidance on when each applies. The following definitions clarify their roles.
RMS velocity (mm/s or in/s): The standard metric for assessing overall machine severity per ISO 10816 and ISO 20816. It reflects average vibration energy and correlates with structural fatigue risk. Use RMS velocity for trend-based condition assessment and regulatory compliance reporting.
Peak acceleration (g): Captures the maximum instantaneous force, making it sensitive to shock events like a ball striking a spall crater. Peak is most useful in Stage 2 and Stage 3 detection when the fault produces repeated impulses. It is noisier than RMS and not suitable as a primary overall severity metric.
Kurtosis: Quantifies the statistical impulsiveness of the signal, expressed as the fourth statistical moment normalized by the square of the variance. A purely random (Gaussian) signal has kurtosis of 3. Early bearing defects generate repeating impulses that raise kurtosis toward 6 to 8, sometimes higher, while overall amplitude remains unremarkable. As faults progress to Stage 3 and 4, kurtosis can paradoxically decrease because the signal becomes uniformly chaotic rather than periodically impulsive.
Crest factor: Peak divided by RMS. Tracks the growing impulsiveness of a signal in proportion to its background level. A rising crest factor alongside a stable RMS often indicates the earliest detectable bearing degradation.
Envelope (demodulation) analysis: A processing technique that extracts the amplitude modulation of a high-frequency carrier signal to reveal low-frequency bearing fault frequencies. Essential for confirming which specific bearing element (inner race, outer race, rolling element, cage) is damaged at Stages 2 and 3.
ISO 10816 and ISO 20816 Bearing Vibration Severity
ISO 10816 (measurement on non-rotating parts) and its successor ISO 20816 (covering both rotating and non-rotating parts) define four vibration severity zones used in condition-based maintenance decision-making.
- Zone A: Newly commissioned or recently maintained machine in good condition. No action required.
- Zone B: Acceptable for unrestricted long-term operation. Continue routine monitoring.
- Zone C: Unsuitable for continuous long-term operation. Investigate root cause, increase monitoring frequency, plan maintenance within weeks.
- Zone D: Vibration levels capable of causing machine damage. Shutdown and inspect.
Specific RMS velocity thresholds (mm/s) vary by machine class. For small machines up to 15 kW, Zone C begins at approximately 4.5 mm/s and Zone D at 7.1 mm/s. For large machines above 300 kW, Zone C begins at around 7.1 mm/s and Zone D at 11.2 mm/s. Always consult the specific machine class and mounting type in the relevant standard part for precise thresholds.
Common Bearing Failure Causes
Identifying the root cause of a bearing failure is as important as detecting it. Without cause analysis, the same failure mode will repeat.
Lubrication Failure
Inadequate lubrication is the most common bearing failure cause. This includes insufficient lubricant quantity, lubricant degraded by heat or oxidation, wrong lubricant type or viscosity for the application, and grease over-packing that generates heat by churning. Symptoms include early-stage ultrasound elevation, metallic particle contamination in oil samples, and discoloration of the lubricant from heat.
Contamination
Solid particle contamination introduces abrasive material between rolling surfaces, accelerating wear and reducing bearing life exponentially. Even ISO cleanliness class degradation of one or two levels can halve bearing life under heavy load. Water contamination causes corrosive pitting and accelerates lubricant degradation. Contamination is best detected through oil analysis particle count and spectrometric elemental analysis.
Misalignment
Shaft or housing misalignment generates additional radial and axial loads that exceed the bearing's design envelope. This produces characteristic vibration patterns at 1x and 2x rotational frequency in the axial direction. Misalignment is a setup and installation failure mode; it is best prevented through precision alignment at installation and detected through vibration spectrum analysis during routine monitoring.
Overloading and Improper Mounting
Loads exceeding the bearing's dynamic capacity rating cause accelerated fatigue. This can result from process changes that increased flow, pressure, or mass on a machine not originally sized for the new duty. Improper mounting practices, such as driving forces through the wrong bearing ring or using damaged tooling, create immediate sub-surface damage that accelerates the progression through the four failure stages.
Electrical Current Damage
Variable frequency drives (VFDs) can induce shaft currents that discharge through bearing contact surfaces, creating pitting patterns called fluting. Fluted bearings produce a distinctive noise, elevated high-frequency vibration, and characteristic pitting visible on raceways at inspection. Prevention requires insulated bearings, shaft grounding rings, or insulated couplings. Detection relies on vibration spectrum analysis and direct inspection.
Implementing a Bearing Monitoring Program
A well-structured bearing monitoring program follows a four-part framework.
Asset Criticality Classification
Not every bearing warrants continuous monitoring. Asset criticality analysis ranks machines by consequence of failure on production, safety, and cost. Critical assets receive online continuous monitoring with automated alerting. Important assets receive periodic measurement on a defined route schedule. Non-critical assets run to failure or receive periodic manual checks.
Measurement Point Standardization
Bearing measurements must be taken at the same physical location, orientation, and sensor attachment method on every measurement cycle. Even small variations in measurement point can mask real trends or create false alarms. Mark measurement points permanently. Use magnetic bases or stud-mounted accelerometers for critical assets.
Baseline Establishment
Establish baseline measurements on new or freshly maintained bearings under normal load and speed. Record operating conditions (speed, load, temperature) at the time of measurement. All subsequent trend analysis compares against these baselines. A bearing that enters service already in poor condition requires earlier baseline re-establishment after correction.
Alert Threshold Setting and Escalation
Set initial alert thresholds conservatively at two to three times the established baseline, then adjust based on observed variability on each asset. Use a two-tier alert structure: a warning level that triggers increased monitoring frequency and investigation, and an alarm level that triggers maintenance scheduling or immediate shutdown evaluation. Alert thresholds should be specific to each asset, not generic site-wide values.
The Bottom Line
Bearing monitoring is not an optional add-on to a maintenance program. It is the most direct means of preventing the single most common cause of rotating equipment failure. The four-stage failure model gives maintenance teams a structured framework for interpreting measurement data, assigning urgency, and planning repairs at the right time. Combining vibration analysis, ultrasound, temperature, and oil analysis covers the full progression from sub-surface crack initiation to imminent failure, ensuring no fault stage goes undetected. Teams that implement structured bearing monitoring consistently reduce emergency repairs, extend bearing service life, and shift maintenance from reactive firefighting to planned, controlled activity. The investment in sensors, software, and technician training pays back through avoided failures, reduced parts consumption, and production hours protected.
Monitor Bearing Health Before Problems Become Failures
Tractian's condition monitoring platform combines continuous vibration, ultrasound, and temperature analysis on a single sensor, giving maintenance teams automatic bearing fault detection, stage classification, and actionable alerts without manual data collection.
See How Tractian Monitors BearingsFrequently Asked Questions
What is bearing monitoring?
Bearing monitoring is the continuous or periodic measurement of physical signals from rotating equipment bearings to detect degradation before it becomes failure. It uses vibration, ultrasound, temperature, and oil analysis data to track bearing health and trigger maintenance at the right time.
What are the four stages of bearing failure?
Bearing failure progresses through four stages. Stage 1 is detectable only by ultrasound, with sub-surface stress cracks forming but no visible damage. Stage 2 produces high-frequency vibration spikes detectable in the 500 Hz to 2 kHz range, though the bearing may still run smoothly. Stage 3 shows clear vibration sidebands, rising overall vibration levels, and audible noise. Stage 4 is near-catastrophic failure, with severe vibration, rapid temperature rise, and risk of seizing within hours.
What is kurtosis in bearing monitoring?
Kurtosis is a statistical measure of the impulsiveness of a vibration signal. A healthy bearing has a kurtosis value near 3. As defects develop, the signal becomes spikier and kurtosis rises. Values above 6 typically indicate significant bearing damage. Kurtosis is particularly sensitive to early-stage defects because it responds to the shape of the signal distribution rather than its overall amplitude.
How does ISO 10816 apply to bearing monitoring?
ISO 10816 (now largely superseded by ISO 20816 for new installations) defines vibration severity zones for rotating machinery measured at the bearing housing. Zone A is new machinery in good condition. Zone B is acceptable for long-term operation. Zone C requires short-term monitoring and intervention planning. Zone D indicates immediate shutdown risk. These thresholds are expressed in RMS velocity (mm/s) and vary by machine power class and mounting type.
What causes most bearing failures?
Studies consistently attribute roughly 80% of premature bearing failures to four root causes: inadequate or contaminated lubrication (approximately 40%), improper installation or handling (around 30%), overloading or misapplication (roughly 10%), and fatigue from normal or accelerated wear cycles (around 20%). Electrical current damage from VFDs is an increasingly common fifth cause in modern facilities.
What is the difference between RMS and peak vibration measurements for bearings?
RMS (root mean square) velocity measures the average energy in a vibration signal and correlates well with overall machine condition and bearing wear. Peak measurements capture the highest instantaneous amplitude and are more sensitive to impact events like spalling or crack propagation. For early-stage bearing fault detection, peak and crest factor are more useful. For trend-based condition assessment and ISO severity rating, RMS is the standard metric.
When should ultrasound be used instead of vibration analysis for bearings?
Ultrasound monitoring is most effective for detecting Stage 1 bearing defects that have not yet produced measurable changes in the vibration spectrum. It is also the preferred technique for very slow-speed bearings (below 100 RPM) where vibration analysis lacks sensitivity, for detecting lubrication deficiency in real time, and for pinpointing the exact location of a fault on multi-bearing machines. Vibration analysis is preferred for trending progression through Stages 2, 3, and 4.
Related terms
Unplanned Maintenance
Unplanned maintenance is any corrective action performed outside a scheduled plan. Learn causes, costs, prevention strategies, and key metrics like MTBF, MTTR, and OEE.
Total Effective Equipment Performance (TEEP)
TEEP measures how effectively equipment runs against total calendar time (24/7/365), not just scheduled shifts. Learn the formula, OEE vs TEEP, and how to improve utilization.
Asset Utilization: Definition, Formula, and How to Improve It
Asset utilization measures how effectively a company uses its assets to generate revenue. Learn the formula, KPIs, benchmarks, and strategies to improve equipment utilization.
Defect Density: Definition, Formula, and How to Reduce It
Defect density measures total defects relative to a defined inspection basis. Learn the formula, types of defects, SPC tools, and how to reduce defect density in manufacturing.
Maintenance Backlog: Definition, Formula, and How to Manage It
A maintenance backlog is the sum of all pending work orders a team must complete. Learn how to prioritize, calculate, and manage it using a CMMS.